Abstract
In this paper, a procedure is proposed to decompose large scale systems with quadratic performance index into a number of interconnected subsystems controlled by a decentralized multilevel scheme. And using it, the interactions of local controllers are dealt with in the same way as the inherent interconnections among states without the assumption that the original system have been (translated into) an input-decentralized one. Furthermore a specific way is shown to determine higher level control operation so as to ensure the local autonomy and improve the overall performance as much as possible. Finally, a numerical example is given to illustrate these results.
